Edison exit polls is the latest iteration.
Since 2000, exit polls have been reporting higher democratic totals than show up in the election results.
In 2000 or 2004, they made up a phrase to explain it Reluctant Bush Responders.
They started adjusting the polls more heavily to represent the percentage of respondents in each demographic needed to EQUAL the results. WEIGHTED polls are what we the public are allowed to see.
They pretend that the purpose of polling is to give INSIGHT on different demographics. Basically so politicians can use it in the future to know what groups or issues they want to target.
The purpose of polling in the 1970s and 1980s and 1990s was to predict quite accurately the results.
I used to watch them as a kid, and saw that they almost always came in within 0.2 % or maybe 0.4%.
But now, you cant even find the unadulterated polls. They are more complicated because of absentee and early voting but yes, they do they have some people standing outside the polling places.
Im not making this up. Look for Edison exit polls and the fine print tells you about the weighting.
I think if you are a statistician professor or something, it is possible to buy pieces of the raw data for small regions, at a price. But you cant buy it and publish it, and who would know what to believe anyway?